3
\$\begingroup\$

Many manufacturers suggest reforming an electrolytic capacitor after it has been stored for a long period. For example, Vishay describes this process:

enter image description here

(source)

An increase in ESR could happen because of oxidation layer depletion. Over time, the dielectric oxide layer within the capacitor can slowly deteriorate, especially if not under voltage. This depletion can lead to an increase in ESR, as the capacitor loses some of its ability to store and release charge efficiently.

There are several sources from manufacturers describing how reforming should be done and general information about how this whole thing works inside an electrolytic capacitor:

https://www.vishay.com/docs/28356/alucapsintrobcc.pdf

https://www.mouser.com/pdfDocs/UCC_ElectrolyticCapacitorTechnicalNotes.pdf

https://www.cde.com/resources/technical-papers/AEappGuide.pdf

https://www.kemet.com/en/us/technical-resources/aluminum-electrolytics-and-applied-power.html

The method I've found most common among them all and that I've used consistently over the years is to add a 1k resistor in series with the capacitor being reformed, apply the full rated capacitor voltage and wait for 2-4 hours.

Then measure the capacitor leak current. An easy way is to measure the differential voltage across the 1k resistor and divide it by 1k. Also check its capacitance and ESR. If these three parameters are within spec, it was a successful procedure and the cap can be used again.

Note that this may not be suitable for high-voltage electrolytic capacitors.

I have two questions:

  1. Some people, mostly "experts" on forums insist on bringing the voltage slowly up. Also, the only manufacturer that suggests this that I've seen is Illionois Capacitors:

enter image description here

Unfortunatelly I could not find the source of this document, I just have it as a local file.

Does anyone know the details of the chemistry involved and what the differences are between slowly bring the voltage up and applying the full rated voltage at once?

  1. Recently I reformed a Nichicon 470uF 63V electrolytic capacitor that was brand new about 7 years ago and was left seated, unpowered (ever). Its capacitance and ESR were within spec before I reformed it.

It turns out that as soon as I applied its rated voltage, the current was in the microamps region and it stayed there for the next 2 hours.

Is the expected reforming chemistry actually happening with currents so low, or can one conclude that, in a case like this, there is no need to reform it and it can be put to use already? Put another way, is there a scenario in which it was left unpowered for over an year, then its capacitance and ESR are within specs, and it is still advisable to reform it?

Adding to the original post, I'll mention that I've used 1000uF and 4700uF high quality capacitors (Nichicon) that were stored for about 2 years, as simple capacitors after a diode bridge. Nothing else connected to the circuit. A few of them exploded right when the circuit was powered on.

This source discusses issues, among other things, regarding the anodized film and failure to self healing:

If the leakage current is too large, in severe cases, it will cause fatal failures such as operation and breakdown of the pressure release device (explosion valve) of the aluminum electrolytic capacitor.

I believe we can extrapolate this condition to the electrolyte degrading the anode oxide layer due to long-term storage. Since the dielectric strength has been reduced by the electrolyte, the capacitor has a higher leakage current than originally. This condition could explain the catastrophic failure of the above-mentioned capacitors, as the inrush current in this case would trigger irreversible damage.

Also this source discusses the storage and reforming process:

enter image description here

This reinforces my hypothesis that current is needed for the self-healing effect to take place, it is actually the V * I energy that causes the reverse electrochemical reaction that regenerates the natural oxidation of the aluminum.

In this case, if one starts reforming a capacitor and during the first seconds or minutes the leakage current - that is the only current taking place - is constant and below specification, there is no need to do the full 2 to 4 hours of reforming.

Another reason I am asking this is because I have thousands of high quality capacitors (think Nichicon, Kemet, Cornell Dubilier, Vishay) that have been stored for over 5 years. When I deploy them (I won't send them back to the factory), it will be good to know if I can stop the process earlier once there is no more excessive leakage current without compromising the purpose of the reforming procedure.

\$\endgroup\$
6
  • \$\begingroup\$ I don't. But Heathkit used to make an IT-28 used for this purpose (up to 600 V, if memory serves.) Might look at its schematic to try answering questions. \$\endgroup\$ Commented Apr 11 at 21:01
  • \$\begingroup\$ Bit surprised that deterioriation of the oxide layer would increase ESR – I can see how it'd increase leakage and decrease capacity, but not how it would increase ESR. Do you have any source where I can read about that? \$\endgroup\$ Commented Apr 11 at 21:11
  • \$\begingroup\$ 7 years seems rather long for an electrolytic capacitor to be unused, but If the capacitor has low leakage at rated voltage, and is within spec for capacitance, then I don't see a need to reform it. Reforming it may possibly lower the leakage even more, but there is a slight risk that you will increase the dielectric thickness, and consequently reduce the capacitance. I, however, can't give you an evaluation of that risk. \$\endgroup\$ Commented Apr 11 at 22:44
  • \$\begingroup\$ @MarcusMüller I have read about it, but when I looked for a source, it scaped me. You can find a start here mouser.com/pdfdocs/ELNAReliabilityAlumElecCaps.pdf , relating an increase in ESR with the dissipation of electrolytes and increase in the tangent of the loss. I admit more sources would be better. \$\endgroup\$
    – tfm
    Commented Apr 12 at 1:39
  • \$\begingroup\$ @tfm yep, but that document alos doesn't have increased ESR as failure mode; sure, they list worsening, tan delta, but that's due to decreased capacity, not due to increased ESR. \$\endgroup\$ Commented Apr 12 at 8:51

1 Answer 1

0
\$\begingroup\$

Reforming an electrolyctic capacitor and stopping the procedure before the suggested time

In this case, if one starts reforming a capacitor and during the first seconds or minutes the leakage current - that is the only current taking place - is constant and below specification, there is no need to do the full 2 to 4 hours of reforming.

Let's call your method "quick reforming". I would advise against it, unless your capacitor is put in active use soon after the procedure. If a new electrolytic capacitor degrades in, say, 2 years of non-use, I would suspect that a "quick reformed" capacitor would degrade in a much shorter time.

According to Nippon Chemi-Con Corporation, the dielectric layer of an aluminum electrolytic capacitor may be 1.1 to 1.5 nm thick per withstand volt. When the dielectric layer is chemically attacked by the electrolyte, its thickness will decrease unevenly. Areas where the dielectric is especially thin, or non-existent, will have greater leakage current density than those areas which are thicker. Reforming current will build up these areas faster because they have higher current density. Because "filling in the holes" proceeds quickly, the leakage current will drop quickly. This is what is happening with your "quick reforming". But the quick falling of leakage current does not mean that the dielectric has achieved a uniform thickness. The longer a reforming voltage is applied to a capacitor, the more uniform the thickness will become, because the current is distributed preferentially to the areas of thinner dielectric. A capacitor with a more un-even dielectric thickness will degrade from no-use conditions faster than one with a less un-even dielectric thickness.

So, if you do a quick reform, and then immediately put the capacitor to use, the reform may well "stick". (This is especially true if the capacitor in use is exposed to near it's rated voltage). But if you do a quick reform, and then let the capacitor sit, you may find that there was a rebound in the leakage current.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.